The present invention relates to active-active high availability (HA) storage systems and remote copy systems and, more particularly, to method and apparatus to virtualize remote copy pair in three data center configuration.
U.S. Pat. No. 7,529,950 discloses remote copy processing and, more specifically, journal technology to assure data consistency in a remote copy system. US 2011/0066801 discloses a cluster storage system having quorum disk. The system is an active-active HA storage system. When the communication path failure between the storage systems occurs, it determines which storage system continues processing by using lock disk (i.e., quorum disk). U.S. Pat. No. 8,578,204 discloses a cluster storage system having witness software, which is used instead of the quorum disk. In the case of the quorum disk, the storage systems execute arbitration processing. In the case of the witness software, the witness software accesses the storage systems and determines which storage system continues processing. The basic theory or concept, however, is same. U.S. Pat. No. 7,373,470 discloses a three data center remote copy system as a multi-target remote copy system. The first volume in first storage system is copied to the second volume in second storage system synchronously. The first volume is copied to the third volume in third storage system asynchronously. Each volume has a different volume ID. The server storing the data in the storage recognizes the three storage systems.
For remote copy of the data stored in an active-active HA storage system, the physical system configuration will be the same as the multi-target remote copy system. The first volume and second volume have the same ID in the active-active HA storage system and the server recognizes them as one volume. It is called virtual volume. The server should recognize one remote copy pair from the virtual volume to copy destination volume in third storage system. However, there are two remote copy pairs physically. Hence, with different remote copy pair IDs, two remote copy pair statuses will be reported to the server. For the server, clustering software such as a SRM (Site Recovery Manager) or traditional script cannot control and operate the remote copy system correctly.
The traditional multi-target remote copy system does not provide active-active features. The server can have access to only a first volume. When the first volume or first storage system failure occurs, the server, clustering software or script recognizes it and executes fail-over processing to the second storage system. The server can change the remote copy status between the second storage system and the third storage system from standby to active. However, in an active-active configuration, IO (Input/Output) from the server is not rejected after the first storage system failure. The IO is processed in the second storage system. Thus, there is no chance to change the remote copy pair status because the server cannot recognize the failure. This causes suspension of the remote copy data protection.